So, he never actually said it was a cumulative failure curve. I'll admit a graph of the probability of failure per year (non-cumulative) may be less expected, but doesn't necessarily indicate an error on the authors part.
And the bit about his estimates of the number of conspirators in the NSA.. he says the exact same thing: "In the PRISM case, the figure of 30,000 comes from total NSA staff. In reality, the proportion of those employed would would have knowledge of this program would likely be a lot less but we take the upper bound figure to minimize the estimate of p."
And the bit about his estimates of the number of conspirators in the NSA.. he says the exact same thing: "In the PRISM case, the figure of 30,000 comes from total NSA staff. In reality, the proportion of those employed would would have knowledge of this program would likely be a lot less but we take the upper bound figure to minimize the estimate of p."